Establishment of Tumor Immune Microenvironment Classification Model to Select Patients Sensitive to Immunotherapy

We thank the Editor for the kind letter and valuable comments concerning our article entitled “Exploiting Tumor Immune Microenvironment to Predict Response to Immunotherapy Plus Chemotherapy in NSCLC.” All authors have seriously discussed all these comments. As was proposed in the article by Kim et al.,1Kim T.K. Vandsemb E.N. Herbst R.S. Chen L. Adaptive immune resistance at the tumour site: mechanisms and therapeutic opportunities.Nat Rev Drug Discov. 2022; 21: 529-540Crossref PubMed Scopus (84) Google Scholar advanced human cancer uses several mechanisms of adaptive immune resistance to escape from immune surveillance, including activation of the programmed cell death protein-1/programmed death-ligand 1 (PD-1/PD-L1) pathway and exclusion of tumor-infiltrating lymphocytes. In theory, PD-L1 expression and immune infiltration represent different aspects of the tumor microenvironment. Both are critical for understanding the immune response against the tumor, and their combined use could provide a more comprehensive view. Considering the highest predictive values of PD-L1 mRNA expression and Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data (ESTIMATE) immune score, we chose them for our tumor immune microenvironment (TIME) classification model and effectively predicted survival benefits of immunotherapy in a randomized controlled NSCLC phase 3 study. In the letter, it was suggested that the collinearity between the ESTIMATE immune score and PD-L1 mRNA expression might affect the predictive performance of the TIME classification model. To confirm our conclusion, we also constructed other models on the basis of PD-L1 tumor proportional score (TPS) and CD8 T cells in Supplementary Figure 7A to H. In the model defined by PD-L1 mRNA and CD8+ T cell, consistent results were observed. Only tumors with high PD-L1 expression and high immune infiltration achieved the best survival outcomes. However, in the model, defined by PD-L1 TPS equal to 50% and ESTIMATE immune score, tumors with high immune scores achieved longer survival benefits no matter the levels of PD-L1 TPS expression. A reasonable explanation is that part of tumors with PD-L1 TPS less than 50% and high immune infiltration can also respond to immunotherapy. Therefore, although PD-L1 expression in tumor cells is less likely to be correlated with ESTIMATE score, we still believed our model on the basis of PD-L1 mRNA and ESTIMATE score represents the most predictive combination. In contrast, the correlations between PD-L1 expression and immune infiltration are complex. Previous studies proved that tumors with high tumor mutational burden had also increased the proportion of immune cells and total PD-L1–positive cells in NSCLC, achieving improved clinical outcomes of PD-L1 blockade.2Ricciuti B. Wang X. Alessi J.V. et al.Association of high tumor mutation burden in non-small cell lung cancers with increased immune infiltration and improved clinical outcomes of PD-L1 blockade across PD-L1 expression levels.JAMA Oncol. 2022; 8: 1160-1168Crossref PubMed Scopus (64) Google Scholar Other studies also suggested common targets of immunotherapies such as programmed cell death protein 1, PD-L1, LAG3, and CTLA4, which were up-regulated in the high-infiltration group.3Yan L. Song X. Yang G. Zou L. Zhu Y. Wang X. Identification and validation of immune infiltration phenotypes in laryngeal squamous cell carcinoma by integrative multi-omics analysis.Front Immunol. 2022; 13843467Google Scholar Considering the confounding associations between PD-L1 expression and immune infiltration, the collinearity between the ESTIMATE immune score and PD-L1 mRNA might not influence the predictive performance of our model strongly. According to the previous study, MHC class II molecules are predominantly expressed by professional antigen-presenting cells such as dendritic cells, B cells, and macrophages, and primarily present exogenously derived peptide antigens to CD4+ T cells.4Axelrod M.L. Cook R.S. Johnson D.B. Balko J.M. Biological consequences of MHC-II expression by tumor cells in cancer.Clin Cancer Res. 2019; 25: 2392-2402Crossref PubMed Scopus (208) Google Scholar In our research, we also proved that the MHC class II signature was enriched in patients with high immune infiltration no matter the level of PD-L1 expression. Therefore, we think that MHC class II antigen presentation is strongly correlated with the status of immune infiltration. To simplify our classification model for clinical practice, we did not include antigen-presenting pathway or CIITA expression in our model. Further studies are warranted to validate the efficacy of our model. In addition, establishing more accurate and interpretable predictive models by using comprehensive information from the TIME is a direction worth further exploring in the future. Dongchen Sun, Jiaqing Liu, Li Zhang: Conceptualization, Writing-original draft, Writing-review and editing. The study was funded by the Chinese National Natural Science Foundation Project (grant number 81872499) and China Postdoctoral Science Foundation (grant numbers BX2021393 and 2022M723634).